A retrospective case-control study of 200 male cases of oesophageal cancer and 778 population controls has been carried out in Ille-et-Vilanine (France). The logarithms of the relative risks of developing the disease increase linearly with daily consumption of alcohol and tobacco independently. The combined effect of both fit with a multiplicative model which is proposed. This model could be applicable to other situations. It explains the sex ratio and the urban/rural differences observed in Ille-er-Vilaine. The practical implications for public health purposes are briefly discussed.